mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-13 22:18:36 +08:00
2d0f93a054
* Refactor: Restructure C++ tests for better modularisation of non-shared code Start cleanup of pytest code for C++ tests Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> Clean up names and remove references to test_cpp.py Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> WIP Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> Move multi-GPU code Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> Update doc and try un-waiving Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> * Update multi GPU file check Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> * Address minor multi-GPU setup bug Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com> --------- Signed-off-by: Dom Brown <3886319+DomBrown@users.noreply.github.com>
38 lines
1.1 KiB
Python
38 lines
1.1 KiB
Python
import os as _os
|
|
|
|
import defs.cpp.cpp_common as _cpp
|
|
import pytest
|
|
|
|
|
|
@pytest.mark.parametrize("build_google_tests", ["80", "86", "89", "90"],
|
|
indirect=True)
|
|
@pytest.mark.parametrize("test_group", [
|
|
"batch_manager", "common", "executor", "kernels", "layers", "runtime",
|
|
"thop", "utils"
|
|
])
|
|
def test_unit_tests(build_google_tests, test_group, build_dir, lora_setup):
|
|
|
|
xml_name = f"results-unit-tests-{test_group}.xml"
|
|
|
|
# Discover and run the actual gtests
|
|
ctest_command = [
|
|
"ctest",
|
|
"--output-on-failure",
|
|
"--test-dir",
|
|
f"{build_dir}/tests/unit_tests/{test_group}",
|
|
"--output-junit",
|
|
f"{build_dir}/{xml_name}",
|
|
]
|
|
|
|
parallel = _cpp.default_test_parallel
|
|
if parallel_override := _os.environ.get("LLM_TEST_PARALLEL_OVERRIDE", None):
|
|
parallel = int(parallel_override)
|
|
|
|
cpp_env = {**_os.environ}
|
|
|
|
_cpp.parallel_run_ctest(ctest_command,
|
|
cwd=build_dir,
|
|
env=cpp_env,
|
|
timeout=2700,
|
|
parallel=parallel)
|